Highlights  A calibration transfer optimized NIR method was used to analyze individual seeds.  The protein contents of single rice kernels were evaluated via two NIR methods.  Spectra of individual seeds in different forms can be transferred among each other.

Near-infrared spectroscopy (NIRS) is a rapid, chemical-free, easy to use, and non-destructive analytical technique that has been widely applied to a diverse range of fields. NIRS analyzes the investigated samples through their NIR spectra. However, NIR spectral data are complex and multivariate, so multivariate data analysis methods (chemometrics) are used to interpret and predict the spectra's chemical and physical information. The analysis process is also very complex, involving both data processing and modeling. This paper first introduces basic concepts of NIRS analysis with the aim to show its complexity. The paper then characterizes the NIR spectral data using the "3H" of scientific big data, with the aim to show their challenges. Finally, the paper describes our initial effort on the development of an integrated software system to support efficient real-time NIRS data analysis and management. The paper claims that this development is an important contribution to tackling the challenges of scientific big data. Index Terms-Near-infrared spectroscopy (NIRS), nearinfrared (NIR) spectral big data, software technologies for realtime NIRS big data analysis and management, scientific big data

Simple, rapid, and reliable determination of the biuret content in urea fertilizer is very important for the development of fertilizer industry. A near infrared diffuse reflectance measurement system with a portable spectrometer was developed, in which the reference and dark background spectrum could also be recorded automatically in addition to the absorbance data. The key performances of the proposed NIR system have been tested on urea fertilizer. Numerical experiments showed that the coefficient of determination (R2) of the external validation set was 0.97, with a root mean square error (RMSE) of 0.04%. The ratios of the performance deviation (RPD) value in the calibration and validation sets were 12 and 3.5, respectively. It can be concluded that this NIR system for the determination of biuret content in urea fertilizer may potentially be used as an alternative method to traditional wet chemical methods due to its simplicity, sensitivity, and portability.